A Novel Adaptive RNN-Based Control Strategy for Deep Flux-Weakening Operation of IPMSMs

For the deep flux-weakening speed regulation of interior permanent magnet synchronous motors (IPMSMs), the conventional formula-based tuning method is highly complex. To address this issue, an IPMSM deep flux-weakening control strategy based on an adaptive RNN is proposed. This paper first analyzes traditional deep flux-weakening methods for permanent magnet synchronous motors. Building upon deep flux-weakening control, an adaptive RNN is integrated with the maximum torque per voltage (MTPV) strategy in the deep flux-weakening region. The neural network model is constructed with the input layer comprising the d-axis and q-axis currents from the previous time step, the current motor speed, and the target reference speed, while the output layer provides the required d-axis and q-axis currents for the control system at the current time step. Finally, the algorithm is implemented and validated through a simulation model built in MATLABR2024b/SIMULINK. The simulation results demonstrate that the proposed adaptive RNN-based IPMSM deep flux-weakening control system exhibits improved accuracy and robustness.